If you find a broken link, or have any questions about the Interlocutor, please feel free to contact

Prof. Jim Peterman

Anti-realism and Scepticism, Realism and Common Sense

     The internecine debate among realist and anti-realist philosophers of science has been long and heated.  A growing body of philosophers would also say that it has become intractable and otiose.  The sentiment is familiar: What does science have to gain from the vituperations of the philosophers?  Questioning of this sort is not without historical analogue, of course.  The attainment of a philosophical quietude within which the particle physicist can work unhindered is a goal that would have been perfectly familiar to, say, the Pyrrhonist physician concerned with inducing ataraxia in the distraught patient.  This anti-philosophical attitude is unbecoming in a philosopher:the Pyrrhonist desire to ameliorate inquisitiveness was ill-founded, but no more so than the current desire [1] to mute the philosopher's questioning into the nature of the scientific endeavor.

      My intention is to defend the philosopher's role in discerning things of import in the findings of science--indeed, of discerning whether there are such things to be discerned at all-and in particular I will defend the validity of the metaphysical and epistemological concerns that are debated in the realist/anti-realist controversy.  I do not mean thereby to endorse the interminableness of the debate as it stands.  Nor do I think that the debate is irresolvable.  I will argue that a rational resolution to this debate is to be had and I will identify it as a version of realism.  The anti-realist options, as have to date been propounded,  are poor options, for as I will argue, the route of anti-realism leads ultimately to either a sceptical/solipsistic view of reality in toto, or to considerable internal inconsistency.  The brand of realism I propose, however, has much to recommend it beyond merely a denial of anti-realism: it accords with a sophisticated conception of common sense, it is internally consistent, and it does justice to the findings of science.

     Just as the philosophers who want to abandon the issues of realism and anti-realism have an historical analogue (in Pyrrhonian scepticism), so too do those who take part in the debate on either side.  The history of classical British empiricism is one of purification.  Locke and Berkeley inchoately worked out some of the empiricist tenets, while Hume rendered the bare empiricist skeleton.  Reid [2] saw that Hume's scepticism was inevitable upon the sensationalism/phenomenalism [3] implied by empiricism (as did Hume), and distinguished perception from sensation, resting his theory of knowledge upon the former.  He did not procure certainty, but by positing an innate perceptual faculty, he opened up a field of common sense within which a fallible realism could prosper.  I will argue that empiricistic anti-realism shares immitigable liabilities with sensationalism and that realism maintains the benefits of Reid's commonsensical perceptualism; I believe it's possible to do so without forcing anyone into a procrustean bed.  

Anti-realism

     In talking about anti-realism, I will focus on the sort put forth by Bas van Fraassen.  There are two reasons for this.  One is that van Fraassen is arguably the best known of the empiricistic anti-realists and the one with whom the reader of this essay will probably be most familiar.  The other reason is that van Fraassen's "constructive empiricism" contains the core ideas of what I consider the anti-realist argument, and can thus stand as a paradigm case.  If van Fraassen's anti-realism is faulty, anti-realism (of the type I am concerned to combat) [4] in general is as well. 

     The fundament of van Fraassen's "constructive empiricism" is essentially an axiological claim;the rest issues forth from it.  Namely, it is the denial of the realist axiology that maintains that the goal of science is the progressive acquisition of knowledge about reality&-its entities, their relations, and the laws which govern their behavior-and that this knowledge is explicable not only instrumentally, but also in the deeper sense of understanding  the causes behind phenomena. 

     In place of the aim of acquiring truth or of discovering causal entities, van Fraassen argues that science's aim is to provide an "explanation" of phenomena only insofar as that explanation is understood as a theory (or theories) which are "empirically adequate."  Empirical adequacy need not be coextensive with truth.  In claiming that one's theories are empirically adequate, one is claiming no more than that these theories are amenable to an interpretation such that the theories in question can account for the phenomena they pertain to without inconsistency with the phenomena.  As van Fraassen says:

A theory is empirically adequate exactly if what it says about the observable things and events in this world are true,exactly if it "saves the phenomena."  A little more precisely: Such a theory has at least one model that all the actual phenomena fit inside. (van Fraassen, 1982)

 In other words, theories "explain" the phenomena in the sense that the formalism of a theory can underpin the phenomena in a manner that is not contradicted by the phenomena:it "saves the phenomena" by "explaining" them in this weak sense, without the requisition of truth about reality being attached to the theories.

     The whole argument for empirical adequacy seems to be a sort of apotropaic device to ward off what he takes to be the hubris of those realists who want to get at the causal processes behind the phenomena.  Van Fraassen is simply asserting his pragmatism here.  We should be wary of this, however, for at least two reasons.  The first is that if scientific axiology is purely pragmatic, "thoroughly Heraclitean" as Larry Laudan has it, then science is in danger of becoming strikingly unscientific, as we conceive of the word.  For pragmatic goals are highly relative to who is practicing science.  Creationist "science" and the Soviet genetics fiascoes under Lysenko point out the way in which processes of scientific inquiry may quickly be mangled when placed in the service of particular pragmatic goals, whether religious, political, or otherwise. 

     Of course, van Fraassen would claim that such arguments miss the point.  Science's goal is not vagariously pragmatic, he would say; it is pragmatic in that it should meet the needs of empirical adequacy (beyond which, he avers, we cannot go), and that such needs entail strictures upon science such as fecundity, parsimony, and the rest.  But this argument requires a defense of the pragmatic utility of empirical adequacy itself.  Defense of the pragmatic utility of empirical adequacy begs the question, however: it is rational, claims van Fraassen, to adhere to empirical adequacy alone as a goal of science because it leads to the "success" of science, but the "success" of science is just formulating theories that are empirically adequate.   

     I don't believe that it requires any logic chopping to show the relativism that is attendant upon a pragmatism so deeply imbedded in the heart of scientific axiology.  If science does not supervene upon the actual facts of the matter, then the facts of the matter will not constrain theorizing, and something will have to fill in for it,a pragmatic goal that is  necessarily a relativist goal.  If science is not an explanatory affair, what is the point of doing science?  What is the point of empirical adequacy?  Many in the pragmatist fold would claim that we theorize in service of the prediction, control, or manipulability of nature.  Should this be the case, though, the term "science" would be a misnomer, and those who diligently work toward the creation and rational modification of theories would not deserve the appellation of "scientist"; science would be indistinguishable from technology.

     This pragmatism would have no ground to stand on if the anti-realist contention is not correct.  We should be wary of van Fraassen's pragmatism (reason two) because it does not have this ground to stand on.  It is at this point that the congruence between phenomenalism and empiricistic anti-realism becomes instructive.  Hume had read Bayle and Locke profitably;he knew the inadequacies of Descartes' attempts to prove the existence of external objects and rejected those rationalistic attempts.  Hume argued influentially against certainty about causal connections, and he argued that all that can ever be known with certainty are the fleeting sensations (Hume called them "perceptions") of the moment.  Hume did away with certainty well before Peirce epigrammatized its death.  The epistemological solipsism that was a result fazed some-Hume was not among them-but one of its most perduring results was the widespread abandonment of certainty.  Many philosophers embraced a fallibilist common sense about the existence of external objects; Reid was an early instance.  The problem today is not as pressing as it was once thought to be.  Many philosophers either endorse or tacitly accept some manner of common sense as a matter of course.  This puts the empiricistic anti-realists in a curious position.  For they are no exception to this commonsensical acceptance of the existence of external objects, save one qualification: those entities that are "observable in principle" are accepted with no quarrel, but they deny the rationality of credence in putative entities that are "unobservable in principle." 

     What gives rise to this dichotomy?  Anti-realists need it in order to be simultaneously anti-realists about what hides behind phenomena and non-solipsists about the phenomena themselves.  They thus require a viable, non-arbitrary distinction between entities that are "unobservable in principle" and entities that are "observable in principle."  As Ernan McMullin has said: "Since entities on the one side of the line are ontologically respectable and those on the other are not, it is altogether crucial that there be some way not only to draw the distinction but also to confer on it the significance that van Fraassen attributes to it" (McMullin, 1984).  This distinction is the by now infamous Observation/Theory (O/T) distinction. [5]   The problem for empiricistic anti-realists is that this line is untenable, as I will argue.

     The locus classicus of arguments against this distinction is Grover Maxwell's 1962 paper "The Ontological Status of Theoretical Entities."  There, he presents a sort of Zeno's paradoxes redux in which he demonstrates the continuum of observation from the level of naked vision to the microscopic level.  Any point along the continuum at which we may choose to draw a distinction between what is observable and what is unobservable will be arbitrary.  Van Fraassen draws the line, fuzzily, on the basis of human perceptual faculties, i.e., the five senses. [6]   This leads to odd consequences: an entity's ontological status depends upon whether or not it can be detected by human beings with their flesh and blood sense organs.  If in several millennia the human species has evolved [7] sense organs that can detect new aspects of the world, then are these aspects of the world therefore "created" by the evolution of the newly acquired sense organs, or "discovered" by them?  Van Fraassen's anti-realist would have to maintain the former, leaving him or her open to the familiar criticism that anti-realists confound the objects of knowledge with the knowledge of objects.  This is also a manifestation of the typical pragmatist failure to distinguish between reasons for holding a belief to be true and reasons why the belief holds true.         

     Laying aside the reductio ad absurdum, it seems entirely apposite to ask whether the argument for  the unobservability of particular types of entities (those unobservable by unaided human natural sense organs) is well crafted.  If it is well crafted, it entails that our natural sense organs are qualitatively different from the artificial sense mechanisms that we have devised to detect these sorts of entities.  For if they are qualitatively similar, then there is no reason to give credence to the results of one sort of detection and withhold it from the results of the other sort.  By qualitatively similar, I mean to say that they operate on much the same physical/physiological principles.  Indeed, it is the case that they do operate on similar principles.  Many sorts of entities that van Fraassen would deny positive ontological status are regularly observed, not with unmediated human eyes or other natural sense apparati, but with sense apparati that humans have developed by clever artifice and lots of scientific theory:cloud chambers, scanning electron microscopes, radio telescopes, and so forth.  We understand at least as much about many of these apparati as we do about our natural sense organs. [8]   An argument is often made that because the instruments that we construct which are frequently used in scientific inquiry involve numerous layers of theory, therefore Quinean underdetermination should vitiate our trust in them.  What is commonly overlooked is that our natural sense organs involve numerous layers of theory also: optical, biochemical, neurological, psychological.  The point here is that our knowing of entities is separable from our perceiving of them.  We are surely psychologically inclined to trust a priori our natural senses more than the artificial apparati, but we are not really rationally justified in doing so, at least insofar as we do understand the artificial sense instruments to a degree equivalent to our natural senses. [9]         

     The slide through idealism to sensationalism is inevitable.  If we deny commonsensical reality to all of the objects of an external world, we circle back to Hume.  If on the other hand we grant positive ontological status to the objects of our experience, we must grant it to all of the objects of our experience;this includes entities that are unobservable by our unaugmented sense organs.  The distinction between entities that are "observable in principle" and entities that are "unobservable in principle" is an ontological Maginot Line.  If we desire to hold it, come what may, we do so only by brute assertion and at the expense of our logical consistency.

The Leap out of Solipsism[10]

     At the base of the realism vs. anti-realism debate is a choice, as radical as Kierkegaard's "leap of faith," that transcends the debate.  It is the choice every agent makes between solipsism and common sense.  Of course, the odds are innately in favor of commonsensical realism: humans have a disposition to believe that their quotidian existence is composed of real objects.  Only the deranged, in fact, go about actively doubting the mind-independent existence of every little thing.  Nature selects against those who do, as they tend to walk off of cliffs and fondle serpents and the like.  But insofar as one is thinking philosophically, one chooses a direction (be it the methodological doubt of Descartes, the therapeutic doubt of Sextus Empiricus, or the common sense philosophy of G.E. Moore) and works out the implications.  It is not that there are no rational factors involved in choosing one side or the other, it is simply that once the die has been cast, so to speak, one must be consistent with the implications of that doctrine.  This is what Hume faithfully did upon empiricist principles, even though he realized the consequences: phenomenalism and epistemological solipsism.  This is where the empiricistic anti-realists equivocate.  They want the parsimony and humility of scepticism, and they want the constructive possibilities of common sense.  The result is a chimera.

     I advocate a leap out of solipsism.  But I haven't, as yet, proposed a substantive landing spot for such a leap.  Of course, the traditional dichotomy of realism vs. anti-realism suggests a disjunction, and my arguments against anti-realism suggest a disjunctive syllogism rendering a leap into realism.  And I do maintain just this.  But nobody would be willing to accede to something of this nature upon this logical model.  One may consider a Pyrrhonian type of scepticism, or offer alternative conceptions of anti-realism.  Realism still wants (in my argument thus far) a positive case made for it.  That is what I intend to offer.

Realism

     Scientific realism is the view that 1) it is justifiably believable that the unobservable entities which are described by a theory are real and accurately described by the theory iff it is justifiably believable that the theory's inferences to empirical conclusions are accurate; and 2) acceptance of a theory (even about unobservable entities) just is to believe the above things are justifiable.  An implication of the above two tenets is that what justifies scientific theory acceptance justifies fallible belief in the entities postulated by the theory as well as in the accuracy to truth of the theory. [11]   This conception of realism is robust enough to accommodate important aspects of both theory realism and entity realism, as delineated below.  It is also in accord with common sense.  Before I can even begin making a positive case for this realism, I must respond to anti-realist criticisms.  This is opportune, as I believe that my defense against a particular line of anti-realist attack will lead me directly in to a case for the brand of realism that I propose to defend.

     Realism, at least theory realism, is often oppugned by means of raising difficulties with theories of reference pertaining to the terms that refer to the entities of a theory that a realist asserts are real.  In particular, it is claimed that the reference of these terms cannot remain fixed, or that it cannot be known if they do, across time as a term is emitted from a dying theory and picked up by a successor or rival theory.  This is supposed to cast doubt on the realist contention that scientific theories converge on the truth, becoming true by ever more narrow margins of approximation as scientific inquiry progresses.  This problem is a sub-problem of more general problems with the correspondence theory of truth. 

     A common rejoinder to these arguments is to accept a form of causal theory of reference, as pioneered by Kripke and Putnam.  This is not a strategy without merits, but it is beyond the scope of this essay to analyze it.  I would instead like to point out that my conception of realism avoids these problems, as the terms of the theory do not refer by virtue of their correspondent truth, but by virtue of the justifiable belief in the accuracy to truth of the empirical inferences, which is constitutive of the justifiable belief in the accuracy of the theory's reference and in the entities referred to.  I shall point out why this is so.

     The argument against the transtheoretical reference of terms has at its core an argument from the fact that a view of reality unmediated by our conceptual framework is unattainable and thus precludes the comparison of theory with reality from a point outside of our conceptual scheme.  As Hilary Putnam has it, we have no "View from Nowhere."  It is further claimed that there can therefore be no noncircular grounding of our factual language.  This argument presupposes the transitivity of the relevant inferential chain.  There are, however, forms of inference that do not exemplify the relation of transitivity between their inferential components, as John Post has argued. [12]  Post summarizes the "View from Nowhere" argument:

(1)     On pain of circularity, we cannot use a part of our language or conceptual scheme to ground or justify the whole of it, in the sense of showing that our language or scheme corresponds somehow to the things in themselves.  We cannot use language to get outside language.

(2)     The only way realists justify, or even make sense of, their position would be by stepping outside language or scheme-that is, by way of some unmediated access or View from Nowhere-in order to see that our language or scheme corresponds to the themes in themselves.

(3)     No such immediacy or transcendence is possible.

(4)     Therefore, realism is unjustifiable, indeed senseless.(Post, 1996)

Post argues that assumption (1) above is a false premise.  It presupposes the transitivity of the relevant inferential justifications.  And there are forms of inference that are non-transitive, so the argument above, if it holds, only holds contingently for certain types of inference, and not necessarily for all forms of inference.

     What types of inferences are non-transitive?  One type is the sort of inference that has the goal of inferring general causes/explanations from sets of particular data.  It is known variously as abduction, retroduction, Inference to the Best Explanation (IBE), a "logic of discovery," and it is the form of inference that I contend is at the core of the scientific project.  Charles S. Peirce coined the term "abduction" to characterize a mode of inference distinct from either deduction or traditional induction.  For Peirce, abduction was reasoning from effects to causes.  Science utilizes two forms of inference, on Peirce's view: induction for the discovery of "laws" and abduction for the discovery of "causes." [13]   Abduction thus has essentially this logical form, (A): given law (x)(FxÉGx); from Ga infer Fa.  N. R. Hanson suggested that abduction should be developed into a "logic of discovery," which would be concerned with providing justification for proposing hypotheses rather than with providing a justification for the acceptance of hypotheses. [14]   This conception of abduction fits well with an entity realism such as Ian Hacking's.  For it is Hacking's thesis that science's goal is to discover the entities that are the causes of the phenomena that science is engaged in explicating; its goal is not the justification of the truth of theories.  In other words, theories are "justified" as predictions, but not as true statements about unobservable entities behind phenomena. [15]  

      In what manner, then, does abductive reasoning escape from the transitivity of inferential justification?  I submit this hypothetical situation: 1) x justifies y because y is the best explanation of x; 2) y justifies z because z is the best explanation of y; 3) therefore, if transitivity of this type of inference holds, x must justify z in virtue of z being the best explanation of x.  We know that this can't be the case, however, because of 1), where y is the best explanation of x, unless of course z and y are identical, in which case the inferential chain is only one step, and this instance is moot.  It also entails that y/z would be self-justifying. 

     An example of the above will help clarify this:

     This amply demonstrates the way in which explanatory relations are different from deductive relations: they are not transitive.  The overemphasis on deduction is a remainder from the time when science was cowed by the ferule of logical positivism.   

     Abductive inference is thus seen to be amenable to the discovery of causal entities.  But while the discovery of entities is a worthy goal of science, it can only be but a part of it, and there needs to be room for the justification of the acceptance of theories about those entities.  Abductive inferences should not and need not be limited to the "logic of discovery."  How to conjoin explanation and confirmation in the abductive process?  Three approaches seem amenable to this task. [16]  

     First, abduction is not limited, as enumerative induction is, in the satisfaction of the following conditions:

                  (CE) (Converse Entailment) If H entails E, then E confirms H.

                  (CC) (Converse Consequence) If E confirms H and K entails H, then E confirms K.

(Niiniluoto, 1999)

This is Ilkka Niiniluoto's formulation, deriving from the work of Howard Smokler, and before him Carl Hempel, on qualitative confirmation.  (CE) and (CC) together imply the possibility of long chains of justification, which has obvious benefits in attaching confirmatory power to abductive inferences. 

     However, (CC) at first sight appears to contradict my argument against the transitivity of abductive inferences made earlier.  This is due to the nature of the terminology that Niiniluoto has used in the formulation of the above conditions.  The way they are stated makes it appear that both (CE) and (CC) are "precise reconstructions" of abductive confirmatory relations,not both are.  I have no quarrel with the lineaments of (CE);as stated, it is a paradigm case of abductive inference and justification, and does no harm to the argument from non-transitivity.  But (CC) is deceptive, for it is characteristically deductive rather than abductive.  This may be seen by a reconstruction of (CC):   [(E confirms H) × (H confirms K)] É (E confirms K).  Niiniluoto's (CC) and my didactic (CC) are equivalent, as K's entailment (or more strongly, deductive explanation) of H implies H's confirmation of K.  Hence, the terms of the Converse Consequence are abductive, yet the form is strictly deductive material implication.  The possibility of procuring confirmation of a chain of abductive inferences is maintained in the deductive form of (CC), while my argument for the non-transitivity of abductive explanatory inferences is not impugned.        

     Secondly, abduction can be interpreted as a form of frequentist probabilistic reasoning.  For the abductive form (A), the frequency of Fx being a true explanation of Gx could be any number between zero and one.  This need not dash the hopes of understanding abduction in a probabilistic light.  If the abductive inference is understood as an enthymeme (of the Aristotelian ycos or "probable proposition" type) as Fumerton (1980) has done, then reliable background beliefs may stand as the omitted premise and the probability of a given abductive inference may be calculated.      

     Thirdly, abduction can be fashioned in a Bayesian mold.  Bayesian analysis attempts to calculate the probabilities of causes given effects.  Suppose we have a "law" of the form "H entails E."  Assume further that H is not impossible and E is not necessary.  Then given E (an observed effect), the probability of cause H is greater than the probability of H in the absence of E.  The conditional probability being greater than the probability of the putative cause absent its entailed effect, we say that explanandum and explanans are positively correlated and that therefore positive instances of the explanans confirm the explanandum.  This Bayesian analysis just is a calculation of the probability of an abductive inference of the form (CC), thus an abductive inference carries with it confirmatory power in the form of a Bayesian probability.  Qualitative confirmation, frequentist probability, and Bayesian epistemic probability, then, all seem to offer considerable promise of backing up abduction's power of discovery with confirmatory power. [17]

     Beyond this, abduction may be profitably used to argue for the truth or approximate truth of scientific theories by an application of abductive inference to the success of scientific theories at prediction.  Given the empirical success of scientific theories, an abductive inference to the hypothesis that these theories are at least approximately true may be made.  Abduction thus seems to underpin the "no miracles" or "cosmic coincidence" argument for scientific realism (Boyd, 1984).  

     Not surprisingly, van Fraassen objects to these approaches, and in particular has offered technical counter-arguments to the Bayesian approach. [18]   It is at this juncture that it will be profitable to resume the argument about realism vs. anti-realism in light of the choice of commonsense or sceptical principles.  For van Fraassen's complaint against the Bayesian approach springs from his pragmatic empiricism.  He claims in effect that since the IBE is always made from among a historically available set of explanations, the IBE may simply lead to "the best of a bad lot."  But again, I will argue, van Fraassen makes this claim in accord with his dualistic and untenable conferrence of common sense to only quotidian entities.

     Van Fraassen overlooks or disregards a relationship between abduction and perception that Peirce makes explicit:our perceptual judgements are only an "extreme case of abductive inference."  Peirce proposed this in part to circumvent the potential objection that abduction commits the fallacy of affirming the consequent (which it does).  Abduction is not strictly a logical inference, but may be rationally compelling in certain circumstances (where background evidence is reliable and so forth), of which perception is one. [19]   The coming to know of historical facts, among many other sorts of inference, is likewise abductive in nature.  Take two examples of abductive reasoning:

1)        A woman walks on the shoreline; she looks out over the waves and exclaims, "A trireme plies the sea."

2)        A librarian opens a tome of history; he reads: "A woman walks on the shoreline; she looks out over the waves and exclaims, "A trireme plies the sea.'"

In the first example, abduction is being used.  She perceives a trireme by the process of abduction: what she senses could be interpreted many ways-is it really a trireme, or is it a whale's back, or a mirage, or an hallucination?-;but what she perceives is an inference to the best explanation.  In the second example, the abduction is a little less powerful, but it is distinctly of the abductive form: the phenomenon here is a description of an event in a book; given the available background evidence (other histories, reliability of the source, archaeological evidence, etc.), he draws the best explanatory inference possible:is the history spurious, best consigned to myth, or is it accurate?  In either case, the inference is to the best possible conclusion upon the available evidence. 

     The implications of the validity of abductive reasoning for a scientific realist position are heartening.  The fundamental choice I suggested earlier is reformularizable: on the one hand is phenomenalism, understood as a denial of the causal element in perception; on the other is common sense, understood as the acceptance of the role of abduction in our perceptual faculties.  Anti-realists throw together elements of both choices:they are commonsensical about everyday entities that can be sensed with our natural sense organs, while they are phenomenalistic about those entities that are subsensible by our natural sense organs.  This can only be done at the expense of their logical consistency. 

     Realism, as I outlined it above, has an advantage here.  Realism makes a definitive choice:the choice for common sense about both those entities that we can sense with our bodies, and those that we can only sense with our artificial sense apparati.  It does so on the basis of a fact about our natural perceptual faculties, namely that they are abductive in nature.  What is more, realism as I see it goes beyond our sense apparati altogether, whether natural or constructed, and allows us to perceive entities that are even truly "in principle" directly unobservable.  Thus, the existence of such entities as black holes and gravitational forces can be rationally inferred, though never phenomenally observed.  We do this on the basis of abductive reasoning, which is shared by but is not limited to our sense perceptions.  Abduction unhooks perception from sensation and allows us to penetrate into the causes of things. [20]

     Perhaps that sounds too optimistic for some ears.  The disposition that finds impudent such claims is the same disposition that engenders anti-realism.  But it appears that the anti-realist position is hubristic.  To refuse to grant reality to an aspect of nature on the grounds that have been proposed seems arrogant.  It is as if the anti-realist simply asserts that we cannot know, and therefore concludes that we should not attempt to know.  Realists are quite assiduous in avoiding hubris.  We go out on a limb, surely, but we don't do so with the intention of maintaining despite all evidence that we have done so correctly.  Fallibilism, rather than enforced ignorance, is the great bulwark against hubris.  This I take to be the Socratic wisdom-"I am wise because I know that I don't know"-and not the defeatism of the anti-realists.  Socrates did not forswear the elenchus after this perspicacious insight, nor should we.